اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدLCFI: A Fault Injection Tool for Studying Lossy Compression Error Propagation in HPC Programs
73
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
Error-bounded lossy compression is becoming more and more important to todays extreme-scale HPC applications because of the ever-increasing volume of data generated because it has been widely used in in-situ visualization, data stream intensity reduction, storage reduction, I/O performance improvement, checkpoint/restart acceleration, memory footprint reduction, etc. Although many works have optimized ratio, quality, and performance for different error-bounded lossy compressors, there is none of the existing works attempting to systematically understand the impact of lossy compression errors on HPC application due to error propagation. In this paper, we propose and develop a lossy compression fault injection tool, called LCFI. To the best of our knowledge, this is the first fault injection tool that helps both lossy compressor developers and users to systematically and comprehensively understand the impact of lossy compression errors on HPC programs. The contributions of this work are threefold: (1) We propose an efficient approach to inject lossy compression errors according to a statistical analysis of compression errors for different state-of-the-art compressors. (2) We build a fault injector which is highly applicable, customizable, easy-to-use in generating top-down comprehensive results, and demonstrate the use of LCFI. (3) We evaluate LCFI on four representative HPC benchmarks with different abstracted fault models and make several observations about error propagation and their impacts on program outputs.
قيم البحث
اقرأ أيضاً
Error-bounded lossy compression is a critical technique for significantly reducing scientific data volumes. With ever-emerging heterogeneous high-performance computing (HPC) architecture, GPU-accelerated error-bounded compressors (such as cuSZ+ and c
uZFP) have been developed. However, they suffer from either low performance or low compression ratios. To this end, we propose cuSZ+ to target both high compression ratios and throughputs. We identify that data sparsity and data smoothness are key factors for high compression throughputs. Our key contributions in this work are fourfold: (1) We propose an efficient compression workflow to adaptively perform run-length encoding and/or variable-length encoding. (2) We derive Lorenzo reconstruction in decompression as multidimensional partial-sum computation and propose a fine-grained Lorenzo reconstruction algorithm for GPU architectures. (3) We carefully optimize each of cuSZ+ kernels by leveraging state-of-the-art CUDA parallel primitives. (4) We evaluate cuSZ+ using seven real-world HPC application datasets on V100 and A100 GPUs. Experiments show cuSZ+ improves the compression throughputs and ratios by up to 18.4X and 5.3X, respectively, over cuSZ on the tested datasets.
As machine learning (ML) has seen increasing adoption in safety-critical domains (e.g., autonomous vehicles), the reliability of ML systems has also grown in importance. While prior studies have proposed techniques to enable efficient error-resilienc
e techniques (e.g., selective instruction duplication), a fundamental requirement for realizing these techniques is a detailed understanding of the applications resilience. In this work, we present TensorFI, a high-level fault injection (FI) framework for TensorFlow-based applications. TensorFI is able to inject both hardware and software faults in general TensorFlow programs. TensorFI is a configurable FI tool that is flexible, easy to use, and portable. It can be integrated into existing TensorFlow programs to assess their resilience for different fault types (e.g., faults in particular operators). We use TensorFI to evaluate the resilience of 12 ML programs, including DNNs used in the autonomous vehicle domain. Our tool is publicly available at https://github.com/DependableSystemsLab/TensorFI.
The vast majority of hardware architectures use a carefully timed reference signal to clock their computational logic. However, standard distribution solutions are not fault-tolerant. In this work, we present a simple grid structure as a more reliabl
e clock propagation method and study it by means of simulation experiments. Fault-tolerance is achieved by forwarding clock pulses on arrival of the second of three incoming signals from the previous layer. A key question is how well neighboring grid nodes are synchronized, even without faults. Analyzing the clock skew under typical-case conditions is highly challenging. Because the forwarding mechanism involves taking the median, standard probabilistic tools fail, even when modeling link delays just by unbiased coin flips. Our statistical approach provides substantial evidence that this system performs surprisingly well. Specifically, in an infinitely wide grid of height~$H$, the delay at a pre-selected node exhibits a standard deviation of $O(H^{1/4})$ ($approx 2.7$ link delay uncertainties for $H=2000$) and skew between adjacent nodes of $o(log log H)$ ($approx 0.77$ link delay uncertainties for $H=2000$). We conclude that the proposed system is a very promising clock distribution method. This leads to the open problem of a stochastic explanation of the tight concentration of delays and skews. More generally, we believe that understanding our very simple abstraction of the system is of mathematical interest in its own right.
We present a new algorithm, Fractional Decomposition Tree (FDT) for finding a feasible solution for an integer program (IP) where all variables are binary. FDT runs in polynomial time and is guaranteed to find a feasible integer solution provided the
integrality gap is bounded. The algorithm gives a construction for Carr and Vempalas theorem that any feasible solution to the IPs linear-programming relaxation, when scaled by the instance integrality gap, dominates a convex combination of feasible solutions. FDT is also a tool for studying the integrality gap of IP formulations. We demonstrate that with experiments studying the integrality gap of two problems: optimally augmenting a tree to a 2-edge-connected graph and finding a minimum-cost 2-edge-connected multi-subgraph (2EC). We also give a simplified algorithm, Dom2IP, that more quickly determines if an instance has an unbounded integrality gap. We show that FDTs speed and approximation quality compare well to that of feasibility pump on moderate-sized instances of the vertex cover problem. For a particular set of hard-to-decompose fractional 2EC solutions, FDT always gave a better integer solution than the best previous approximation algorithm (Christofides).
Extreme-scale cosmological simulations have been widely used by todays researchers and scientists on leadership supercomputers. A new generation of error-bounded lossy compressors has been used in workflows to reduce storage requirements and minimize
the impact of throughput limitations while saving large snapshots of high-fidelity data for post-hoc analysis. In this paper, we propose to adaptively provide compression configurations to compute partitions of cosmological simulations with newly designed post-analysis aware rate-quality modeling. The contribution is fourfold: (1) We propose a novel adaptive approach to select feasible error bounds for different partitions, showing the possibility and efficiency of adaptively configuring lossy compression for each partition individually. (2) We build models to estimate the overall loss of post-analysis result due to lossy compression and to estimate compression ratio, based on the property of each partition. (3) We develop an efficient optimization guideline to determine the best-fit configuration of error bounds combination in order to maximize the compression ratio under acceptable post-analysis quality loss. (4) Our approach introduces negligible overheads for feature extraction and error-bound optimization for each partition, enabling post-analysis-aware in situ lossy compression for cosmological simulations. Experiments show that our proposed models are highly accurate and reliable. Our fine-grained adaptive configuration approach improves the compression ratio of up to 73% on the tested datasets with the same post-analysis distortion with only 1% performance overhead.
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
